CCRL Discussion Board

I am looking for drawn positions which are harder for the defender to defend if they have the move than if they have not. 'Difficulty' needs to be quantified in some way, which relates to the above discussion. Is this ... 3k4/1R6/3B4/3K4/8/8/8/r7 ... such a position? There are probably positions whe...

My thought was that the code to choose between drawing moves would be on the search-engine side ... but maybe it's not a clear case. If the engine searches a few ply forward, and probabilities are applied to the leaf-positions of the search-tree (and then backed up as expected-values to indicate the...

I wasn't proposing to store recommended drawing moves in a database.

Much prefer the forward-search with backed-up 'probability of being chosen' instead.

g

Yes, certainly there are ways to rank the drawing moves. First, one could count the percentage of first responses that draw - and pick the move that minimises this. An extreme case of this, if one looks at 'draw studies', is that Black probably chooses a drawing move that forces a unique response fr...

I have updated the 'Chess Endgame Records' dataset/pgn to 'version 7' at http://centaur.reading.ac.uk/34268/ The most significant additions are: 1) A timeline for the history of Endgame Table generation and publication (comments, corrections, additions welcome), 2) Ply-count-critical examples of max...

This is very interesting work featuring two relatively new innovations: - multiple, ply-count-related, DTM50 depths per 'physical' position (pace e.p.), and - the use of the functional programming language HASKELL (and only Joe Hurd has used an FPL (HOL) on EGTs before). Just to correct some minor t...

The latest 7-man DTM statistics double the evidence that log(maxDTM) is increasing linearly with the number of men ... that is to say, that maxDTM is increasing exponentially with the number of men. I give the detail on this in http://centaur.reading.ac.uk/34268/ ... and the graph can also be found ...

I have published a survey of the deepest known endgame positions and lines of play from them. 'Deep' is measured in terms of DTC(onversion), DTM(ate) or DTZ(eroing of the ply-count). The URL is http://centaur.reading.ac.uk/34268/ where a pgn file of positions/lines, and an annotated version of the p...

I wonder if Marc B has the DTC-stats files for KQBNKQB White wins and Black losses.

I would like to know if there are exactly 2 wtm wins in (dtc =) 330.

g

I've responded in part via the Discussion Board's message-facility. I can live with 'Cursed Loss' :-) Ken Thompson tried the 'outcounting' method - presumably used when trying to identify forced-loss positions - and also found it less than worthwhile. The reason for drawing on information from previ...

Thank you for your informative reply. It is really good news that I can set aside my 'April fool' doubts as I hoped. I had just been through a sceptical phase after reading Chessbase's neat April Fool story about FIDE raising the K-factor of the ELO system to 60 (which, at the time of writing, is no...

This is very interesting - and the filesizes quoted for the 3-5-man EGTs (561 MB) is amazing compared with the Nalimov figure of 3718 MB). I must confess that I did wonder - as the announcement was made on April 1st - whether this was an April Fool's joke. However, assuming it is not, may I ask if b...

Emil Vlasak has an interesting article in EG 189 on '7-man alternatives' featuring both 'FinalGen' and the MVL EGT supercomputer initiative at Moscow State University. The latter seems to be making fantastic progress: KQRKQB EGT, e.g., generated in 2'20" is a bit quick. I would be interested in...

I would be interested to hear from those who have used Eiko Bleicher's FREEZER - positions addressed, types of constraint defined, results achieved? This sw certainly deals with 7-man positions with blocked Pawns, e.g., 8/3K1kn1/8/5p2/3b1P2/6Q1/8/8 w - - 0 88 (Anand-Costa, 1988, c/o KM, http://www.c...

Ok, thanks Josh. What I would have done, if I''d had the EGT is to show my engine just one of the twelve component files at a time ... and see whether how it set about 'evaluating' or 'looking up' each position. I have to close down and boot up FRITZ each time to show it a different set of EGTs - bi...

ROOKIE -v- THE BARON: Round 8, Game 29 of the ICGA World Computer Chess Championship 2011, Tilburg, the Netherlands http://www.digitalgametechnology.com/ChessLab/Live/WCCC2011/tfd.htm The game reached this KQPKQP position: 6Q1/3k4/5q2/8/8/1p4P1/8/6K1 b - - 0 65 and THE BARON correctly announced ‘mat...

Does anyone have any information about these?

Thanks, g

Referring to the KQNKRR position 103b above, the only way this fails to be a loss in 101 plies is if White can find an alternative line where it mates or captures (rather than forcing Black to capture a wN with a Rook). For it to be obvious that this option is available, and to back up the depth of ...

The KQNKRR endgame has maxDTM = 174 (wtm and btm) and maxDTC = 152 (wtm and btm). The position r5r1/8/k7/8/8/8/3K4/1Q4N1 b is a maxDTC/M position, and some years ago I followed a maxDTC/M line. After 95 moves by each side, the DTC-minimaxing line and the DTM-minimaxing line parted company. Position ...

As I nearly said, Christoph Wirth generated DTC EGTs with depth in plies. [As usual, I said 'N Wirth' by mistake as I tend to have Nicklaus Wirth's name in my head.] So you won't be the first to measure depth in plies, KK, but you might be the first to measure depth in plies correctly. I think CW in...

I think a lot of complexity about 'storing moves' rather than 'storing depths' is being proposed here. And I don't think it's a good idea anyway. First, it's important to be able to compare a new EGT against, say, a Nalimov EGT, especially if the new EGT is using metric DTM. This cannot be done if '...

Interesting to see mention of Edsger Dijkstra. He visited Cambridge quite often and was, with Tony Hoare, a big influence on my thinking. His EWDs are mostly transcribed at http://www.cs.utexas.edu/users/EWD/ . I don't know of any 'appreciation' of EWD's contribution, but http://www.dijkstrascry.com...

The Schaeffer paper I cited is available at http://webdocs.cs.ualberta.ca/~jonathan/publications/ai_publications/databases10.pdf JS (p8, item 5 on 'Verification') says that a quick scan of an EGT can check for 'internal' self-consistency but that this does not catch all errors. All positions need to...

I don't understand your 'Partitions of 5' table at all ... and I'm not sure why you are doing it. A 'Pawn slice' is a subset of an endgame with all the Pawns on fixed squares. Historically, the word 'tablebase' which most (but not me) use instead of 'table' (i.e., 'EGTB' rather than 'EGT') was coine...

The encoding of depth need not be aligned with the sign-bit of a notional integer of say 8 bits. It needs to cover 'draw', 'index unused' (aka 'broken' in Nalimov terminology), lost in 0-n1 and won in 1-n2. If (n1 + n2 + 3) > 64, Nalimov used a 16-bit integer. The choice of 8 bits or 16 was made by ...